[GRASS-dev] 64bit image issues (WAS: Re: grass-dev Digest, Vol 2, Issue 19)

[Brad Douglas]

On Mon, 2006-06-12 at 12:01 +0100, Glynn Clements wrote:
> Brad Douglas wrote:
> 
> > FYI, it is exiting with status 01 and I can only assume the error has
> > nothing to do with GRASS (looks like the xorg GLX module needs work?):
> 
> Either that, or libGL needs work. What does glxinfo say about the
> server's GLX version?

1.2.

> If the server reports 1.3, it's the server's fault for claiming to
> support GLX 1.3 then complaining about getting GLX 1.3 requests. If it
> reports 1.2, it's the library's fault for trying to send GLX 1.3
> requests to a server which doesn't support them.

Hmm.

> I suppose that the off-screen rendering code should probably have a
> run-time mechanism to disable attempts to create a pbuffer (probably
> for GLX Pixmaps also) in case of buggy OpenGL implementations (which,
> unfortuntately, are quite common).
> 
> If DRI is in use, you could try "export LIBGL_ALWAYS_INDIRECT=1" to
> disable it, in case the problem is with the DRI module.

Doing that makes nviz error out with a GLXBadContext, immediately.  This
is the same problem that Hamish having.

> For now, you can manually undefine HAVE_PBUFFERS (and optionally
> HAVE_PIXMAPS as well) in config.h before compiling NVIZ. That will
> allow you to determine whether there are any other bugs in that code.

#undef HAVE_PBUFFERS errors with GLXBadContext on PixMap (using GLX
1.1).  Adding #undef HAVE_PIXMAPS makes really pretty high resolution
PPMs. :-)


-- 
Brad Douglas <rez touchofmadness com>                      KB8UYR
Address: 37.493,-121.924 / WGS84    National Map Corps #TNMC-3785